I am trying to estimate a measurement model with a latent method factor specified.
I have a 6-factor model, with a method factor (loaded with all the observed variables. The variance of the method factor is constrained to 1, and the correlation between method factor and the 6-factors are constrained to 0.
At first, the model won't converge. Then I increased to 100000 iteration. \ Then I got an error message saying THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING PARAMETER 175.
Parameter 175 is in the psi matrix, covariance between two factor.
Could you help me understand how to approach this issue?
This sounds like a bi-factor model for which one typically has the 6 factors uncorrelated.
J Hu posted on Tuesday, October 02, 2018 - 7:04 am
The constructs of the 6 factors are supposed to be correlated.
But even if I tried to constrained them to be uncorrelated. I still got the same error - standard errors cannot be computed.
I am testing this common method variance (unmeasured common latent factor approach) using guidance provided by 1) Williams, Cote, & Buckley (1989). Lack of method variance in self-reported affect and perceptions at work: Reality or artifact? Journal of Applied Psychology, 74, 462-468. 2) Richardson, Simmering, Sturman (2009). A Tale of three perspective, Organizational Research Methods, 12, 762-800.
Could you help and see if there is anything I could do to fix the models?